Extremely interesting and useful is "The Digital Mathematics Library" by Allyn Jackson in the September 2003 Notices of the AMS (v50i8). As it says, the mathematical literature is one of the best candidates for wholescale retrodigitization primarily because the literature itself has such a long useful life. Like I tell people, 1 + 1 will always = 2. However, the nature of math literature is that it is very diffuse and distributed, and thus any wholescale digitization effort will mirror that. Well, this article is a good overview of the issues and where we're at. The list of active projects alone is terrific. Highly recommended. From SEPW.